Does anybody how much Multi-AZ synchronous replication affects performance? I searched, but I couldn't find any benchmarks. Amazon's FAQ[1] says "You may observe elevated latencies relative to a standard DB Instance deployment in a single Availability Zone as a result of the synchronous data replication performed on your behalf" but they don't quantify it.
Since the hot standby can't serve read requests and is synchronous, I assume something like DRBD is being used on the block level to replicate the data to a separate AZ. I assume AMZN has adequate interconnect bandwidth between their AZs. Gigabit Ethernet has around 100-200μs RTT latency, and then there is the fact that you're writing the data twice...
[1] http://aws.amazon.com/rds/faqs/#39